WO2018197083A1 - Procédé, produit de programme informatique, support lisible par ordinateur, appareil de commande et véhicule comprenant l'appareil de commande pour la détermination d'une manœuvre collective d'au moins deux véhicules - Google Patents

Procédé, produit de programme informatique, support lisible par ordinateur, appareil de commande et véhicule comprenant l'appareil de commande pour la détermination d'une manœuvre collective d'au moins deux véhicules Download PDF

Info

Publication number
WO2018197083A1
WO2018197083A1 PCT/EP2018/055452 EP2018055452W WO2018197083A1 WO 2018197083 A1 WO2018197083 A1 WO 2018197083A1 EP 2018055452 W EP2018055452 W EP 2018055452W WO 2018197083 A1 WO2018197083 A1 WO 2018197083A1
Authority
WO
WIPO (PCT)
Prior art keywords
vehicle
formation
collective
maneuver
maneuvers
Prior art date
Application number
PCT/EP2018/055452
Other languages
German (de)
English (en)
Inventor
Jens Schulz
Kira HIRSENKORN
Julian LÖCHNER
Moritz Werling
Original Assignee
Bayerische Motoren Werke Aktiengesellschaft
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Bayerische Motoren Werke Aktiengesellschaft filed Critical Bayerische Motoren Werke Aktiengesellschaft
Priority to EP18710403.9A priority Critical patent/EP3616184A1/fr
Priority to CN201880009506.5A priority patent/CN110268457B/zh
Publication of WO2018197083A1 publication Critical patent/WO2018197083A1/fr
Priority to US16/657,025 priority patent/US20200050214A1/en

Links

Classifications

    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
    • G05D1/00—Control of position, course or altitude of land, water, air, or space vehicles, e.g. automatic pilot
    • G05D1/02—Control of position or course in two dimensions
    • G05D1/021—Control of position or course in two dimensions specially adapted to land vehicles
    • G05D1/0287—Control of position or course in two dimensions specially adapted to land vehicles involving a plurality of land vehicles, e.g. fleet or convoy travelling
    • G—PHYSICS
    • G08—SIGNALLING
    • G08G—TRAFFIC CONTROL SYSTEMS
    • G08G1/00—Traffic control systems for road vehicles
    • G08G1/16—Anti-collision systems
    • G08G1/167—Driving aids for lane monitoring, lane changing, e.g. blind spot detection
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
    • G05D1/00—Control of position, course or altitude of land, water, air, or space vehicles, e.g. automatic pilot
    • G05D1/02—Control of position or course in two dimensions
    • G05D1/021—Control of position or course in two dimensions specially adapted to land vehicles
    • G05D1/0212—Control of position or course in two dimensions specially adapted to land vehicles with means for defining a desired trajectory
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/50—Context or environment of the image
    • G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
    • G06V20/58—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/50—Context or environment of the image
    • G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
    • G06V20/58—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads
    • G06V20/584—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads of vehicle lights or traffic lights

Definitions

  • the method, computer program product, computer-readable medium, controller and vehicle include the controller for determining a collective maneuver of at least two vehicles
  • the invention relates to a method for determining a collective maneuver of at least two vehicles.
  • the invention further relates to a computer program product, a computer readable medium, a control unit, and a vehicle comprising the control unit for
  • Autonomous vehicles can use a movement prediction to estimate movements of other vehicles and other road users. For example, an currently executed or a future maneuver of a vehicle can be estimated. The amount of possible maneuvers of a vehicle is often determined manually by a manufacturer of the vehicle and provided to the vehicle.
  • the invention is characterized by a method for
  • the method may relate two or more vehicles to each other and determine a collective maneuver for those vehicles.
  • a vehicle may be, for example, an ego vehicle or a vehicle detectable by a sensor system of the ego vehicle.
  • the method may determine one or more collective maneuvers for combinations of vehicles in which the ego vehicle is part of the collective maneuver or the ego vehicle is not part of the collective maneuver, eg if the ego vehicle is a collective maneuver for at least two vehicles determined in an environment of the ego vehicle.
  • the ego vehicle is preferably an autonomously driving vehicle, in particular an autonomously driving vehicle
  • the collective maneuver may include a driving maneuver of the ego vehicle.
  • the method includes receiving a state of a first vehicle, e.g. one
  • the Ego vehicle or other vehicle and a state of at least one other vehicle in an environment of the ego vehicle.
  • the environment may include a road section of the ego vehicle that is detectable by a sensor system of the ego vehicle.
  • the state may include, for example, a position and / or a speed.
  • the state of the first vehicle and / or of the at least one further vehicle can be received by one or more sensors of the ego vehicle, referred to below as sensors of the ego vehicle, and / or control devices of the ego vehicle connected to the sensor system.
  • the method further comprises determining a current formation for the first vehicle and the at least one further vehicle based on the received state of the first vehicle and the received state of the at least one other vehicle in the environment of the ego vehicle, the current formation having a relative order of first vehicle and the at least one further vehicle with respect to a road, in particular one or more lanes of a roadway comprises. Furthermore, the method comprises determining a set of collective maneuvers, in particular a set of possible collective maneuvers based on the current formation, wherein a collective maneuver of the set of collective maneuvers comprises a sequence of formations from the current formation to a final formation.
  • the method further comprises calculating at least one trajectory for the first one
  • Vehicle and at least one trajectory for the at least one further vehicle for a collective maneuver from the set of collective maneuvers are preferably calculated so that the collective maneuver is possible by means of the calculated trajectories, i. can be driven by the vehicles of the collective maneuver.
  • the method includes determining the collective maneuver based on the calculated trajectories for the collective maneuver from the set of collective maneuvers and a movement of the first vehicle and the at least one other vehicle detected by a sensor of the ego vehicle.
  • determining a set of possible collective maneuvers for at least two vehicles it can dynamically respond to a traffic scenario in which For example, one or more, additional vehicles and / or one or more obstacle objects, such as a stationary or parked vehicle, must be considered in the environment of the ego vehicle in the maneuver.
  • the ego vehicle can thus flexibly adapt to new traffic scenarios without having these new traffic scenarios on the market
  • Ego Vietnamese deposited in advance must be deposited in advance. Further, by determining a sequence of formations for each maneuver, efficient maneuver calculation can be ensured. For example, if the method is executed in one-second increments, the ego vehicle may, at each time step, select from the set of collective maneuvers a collective maneuver determined in response to sensed movement of other vehicles. Thus, the ego vehicle can always be an optimal, collective
  • Execute maneuvers for example, to achieve a predetermined goal.
  • the first vehicle may be the ego vehicle, and / or the state of the first vehicle and / or the state of the at least one vehicle may include a preferably track-accurate position of the respective vehicle. This can efficiently determine a collective maneuver for the ego vehicle. Furthermore, a more precise detection of a current traffic situation and thus a more precise detection of a current formation can take place.
  • the method can continue
  • Egohuss include. Further, determining the current formation based on the state of the first vehicle, e.g. of the ego vehicle, the state of the at least one other vehicle, and / or the state of the obstacle objects, and the current formation has a relative order of the first vehicle, e.g. of the ego vehicle, the at least one other vehicle, and the obstacle objects with respect to a roadway.
  • Obstacle objects can be determined efficiently.
  • the sequence of formations from the current formation to the final formation may include a predetermined amount of pairwise, lateral relationships between the first vehicle and the at least one or more further vehicles and / or a predetermined pass order of the first vehicle and the at least one specify one or more other vehicles, and / or may include collective maneuvers from the set of collective maneuvers include a trajectory of the first vehicle and a trajectory of the at least one or more further vehicles, and / or the trajectory of the first vehicle and the trajectory of the at least one or more further vehicles with respect to the current formation and a same end formation be homotopic.
  • Homotop means that the trajectories of all vehicles involved in a maneuver with the current formation as start formation and the same end formation allow a continuous transformation without collision with an obstacle object and / or collision of two vehicles. This can efficiently calculate a collective maneuver.
  • the relative order of the current formation may be accurate in track, and / or the current formation and the end formation may comprise at least one area that is free of vehicles and / or an obstacle object.
  • the maneuver can be precisely recorded by the ego vehicle and calculated efficiently.
  • determining a set of collective maneuvers may comprise generating a tree data structure for formations, wherein the current formation is a root element of the tree data structure. Further, determining a set of collective maneuvers may include calculating a further formation by changing the relative order of the first vehicle, e.g. of the ego vehicle, and / or the at least one or more other vehicles of the starting formation using a predetermined set of discrete motion actions, adding the others
  • Endformation is, include. If the further formation is a final formation, the sequence of formations from the current formation to the final formation can be added as a collective maneuver to the set of collective maneuvers and thus collectively
  • a collective maneuver can be efficiently calculated or determined taking into account the at least one further vehicle and / or one or more obstacle objects.
  • Results of checking whether the further formation is a final formation wherein the determination of the amount of maneuvers is continued until the further formations are completely calculated with respect to the predetermined amount of discrete movements.
  • the trajectory for a maneuver from the set of maneuvers with respect to at least one vehicle-specific parameter and / or with respect to at least one maneuver-specific parameter can be cost-optimal. With this an efficient trajectory can be calculated for every possible maneuver of the ego vehicle.
  • the invention is characterized by a
  • a computer program product for determining a maneuver of an ego vehicle comprising instructions that cause the computer program product to be executed by a computer or controller, the computer, or the controller to perform the method described above.
  • the invention features a computer-readable medium comprising instructions that, when executed on a computer or controller, perform the method described above.
  • the invention is characterized by a control device for determining a maneuver of an ego vehicle, wherein the control device comprises means for carrying out the method described above.
  • the invention is characterized by a vehicle comprising the above-described control unit for determining a maneuver of an ego vehicle.
  • the invention is based on the following considerations:
  • Predicting a movement of vehicles in an environment of an autonomously driving vehicle is a prerequisite for the safe, anticipatory and cooperative driving of autonomous vehicles.
  • intentional movements can not be measured directly. Consequently, it is necessary to make an estimate of movements and / or maneuvers of vehicles by an autonomous vehicle in traffic scenarios in which one or more dependencies between the autonomous vehicle and other, different road users, e.g. Vehicles and / or obstacle objects present.
  • an autonomously driving vehicle can align its own motion planning.
  • the movements of road users are independent of each other. This may mean that the prediction of the movement of the road users and the planning of the movement of the autonomously driving vehicle can not be considered and solved separately. Rather, the prediction of the movement and the planning of the movement are to be solved coherently.
  • maneuvers based on formations are defined, which describe a relative movement of several vehicles in a traffic scenario. For example, linking decisions for or against a maneuver can be made by establishing conditions that are to avoid a collision in a traffic scenario.
  • trajectories for a given maneuver can be scheduled to map human behavior in the traffic scenario.
  • a comparison of the maneuver with observations of a current movement can be carried out and, in the event of a deviation of the planned trajectories from the current movement, a change of the maneuver can be carried out.
  • Bayesian statistics a probability distribution can be derived from possible maneuvers.
  • probability distribution can be used to select and drive a particular maneuver through the ego vehicle.
  • FIG. 2 shows an exemplary traffic scenario and one derived from the traffic scenario.
  • FIG. 3 shows an iteratively expanded tree data structure for identifying maneuvers of the
  • FIG. 5 shows a lateral optimization of the traffic scenario from FIG. 2, FIG.
  • Fig. 6 cost-optimal trajectories for the traffic scenario of Fig. 2, and
  • FIG. 7 shows probabilities for possible maneuvers of the ego vehicle in the traffic scenario of FIG. 2.
  • the maneuver of the ego vehicle is collision-free, ie a collision with one or more further vehicles and / or obstacle objects is avoided.
  • the maneuver may include, for example, overtaking and / or avoiding an obstacle object and / or another vehicle.
  • the maneuver may be passing through an intersection, driving through a roundabout, and / or entering a priority road.
  • the at least one further vehicle may, for example, be Be a vehicle that is manually controlled by a driver or be an autonomous driving vehicle. Communication between the ego vehicle and the other vehicle or vehicles is not necessary.
  • the at least one further vehicle may be, for example, a vehicle that travels on a lane in the opposite direction to the lane of the ego vehicle and / or drives on a lane of a priority road to which the
  • the at least one further vehicle is a vehicle in an environment of the ego vehicle whose movements have an influence on the maneuver of the ego vehicle.
  • the method 100 may first determine a state of the ego vehicle and a state of the at least one other vehicle in the environment, preferably in the immediate vicinity, of the ego vehicle 102.
  • the state of the ego vehicle may determine a position of the ego vehicle
  • the state of the at least one other vehicle may include a position of the at least one vehicle and / or a speed of the at least one vehicle.
  • the position of the ego vehicle and of the at least one further vehicle preferably comprises a longitudinal position and a lateral position of the respective vehicle along a road course or a road section.
  • the speed of a vehicle e.g. The ego vehicle and / or the at least one other vehicle may include a longitudinal speed and a lateral speed.
  • the position e.g. a longitudinal position and / or a lateral position can be determined with respect to a Frenet coordinate system. Furthermore, the position of the
  • the determination 102 of the state may include receiving track-specific map data, for example from a server or from a local data memory integrated in the ego vehicle.
  • a box surrounding the vehicle with a predetermined length and a predetermined width can be determined.
  • the predetermined length of the box and the predetermined width of the box can be determined, for example, depending on the length and width of a vehicle.
  • the box can be mapped to the track accurate map using the position of the vehicle to determine the to-track position of the vehicle.
  • the box surrounding the vehicle is relative to a line, eg to a center line, or a other object of the roadway. This can be a tracking image of the vehicle can be simplified.
  • the method 100 may be a state of one or more
  • the obstacle object may be located in the environment of the ego vehicle and has an influence on the maneuver of the ego vehicle, i. Failure to observe the obstacle object (s) could lead to a collision. Consequently, the or
  • Obstacle objects are considered in the maneuver of the ego vehicle. Analogously to determining the state of his vehicle, a position and optionally a speed of the obstacle object can be determined.
  • the obstacle object is a static object, i. the speed of the obstacle object is zero.
  • a can be
  • Obstacle object surrounding box can be determined and mapped to the exact track map.
  • one or more obstacle objects may be considered in determining a maneuver of the ego vehicle.
  • the maneuver of the ego vehicle can thus be adapted to more complex traffic situations involving one or more obstacle objects.
  • the method 100 may further include a formation for the ego vehicle and the at least one other vehicle based on the determined state of the ego vehicle and the determined state of the at least one other vehicle in the environment of the
  • a formation describes a relative order of objects, e.g. of the ego vehicle and the at least one other vehicle with respect to a road section or a roadway section in a traffic scenario.
  • An object may be a vehicle, e.g. the ego vehicle and / or the at least one more
  • Each formation may include information about a two-dimensional relative position of the objects of the traffic situation. Exact distances and / or lengths between objects can be used in a formation
  • the ego vehicle In order to determine a formation, for a road section, preferably for a road section without intersections, in the environment of the ego vehicle a local, the
  • Road section comprehensive Frenet coordinate system can be defined.
  • the two-dimensional position of one or two a plurality of vehicles, eg, the ego vehicle and the at least one other vehicle, and the two-dimensional position of one or more obstacle objects, if any, on that road section are determined and a projection of the determined positions into the longitudinal dimension is performed.
  • the vehicle (s) and obstacle (s) can be sorted.
  • the vehicle or vehicles and the obstacle or objects are in
  • the sorted order of vehicles and / or obstacle objects can be stored for further calculation.
  • the sorted order can also be extended by a second dimension, in which a track is assigned to each vehicle or obstacle object.
  • the assignment of the track can by means of the track accurate map and the
  • the traffic scenario 200 includes an ego vehicle 202, vehicle A in FIG. 2, a stationary vehicle 204 as an obstacle object, vehicle B in FIG. 2, and another vehicle 206 , Vehicle C in FIG. 2, which accommodates the ego vehicle 202.
  • the ego vehicle may also be the vehicle B or the vehicle C.
  • the vehicle A in FIG. 2 is the ego vehicle 202.
  • the road or the road includes two lanes.
  • the second vehicle 206 and the stationary vehicle 204 are located on a first track.
  • the further vehicle 206 is located on a second track.
  • Also shown in FIG. 2 are the boxes 208, 210, 212 surrounding the vehicles 202, 204, and 206 with which the vehicles 202, 204, 206 are mapped onto the exact map.
  • the longitudinal position of the ego vehicle 202 is s A , the
  • the formation 214 thus comprises a track-accurate position of the vehicles relative to one another.
  • the formation 214 may be represented as a collection of multiple cells. Each cell of the formation 214 is occupied by a maximum of one vehicle or obstacle object. Two vehicles or obstacle objects can not be on the be arranged in the same longitudinal column, ie laterally adjacent cells from an occupied cell are always free. If in the considered road section of the
  • a local Frenet coordinate system comprising the road section and the intersection may also be defined.
  • various road sections of the intersection can be extracted and linkage of the road sections to the intersection derived.
  • first a formation for a first road section and a formation for a second road section crossing the first road section can be determined independently of one another. If a cell of the first formation is also present in the second formation, then that cell is marked as an intersection cell. This is continued until all junction cells are marked.
  • An intersection cell can not be occupied by a vehicle or obstacle object. For example, if an intersection cell is occupied at the time of determining a formation by a vehicle, the vehicle occupying the cell may be set to the next, subsequent, normal cell. It is assumed that it is known which intersection exit a vehicle takes.
  • the method 100 may determine a set of collective maneuvers, based on the determined current formation, also referred to below as start formation 106.
  • the current formation comprises at least one position of the ego vehicle and a position of another vehicle.
  • the set of collective maneuvers may be recalculated in each time step in which the method 100 is executed.
  • a time step may be a fraction of a second, 1, 0 seconds, 1, 1 seconds, 1, 2 seconds, 2 seconds, 3 seconds, and so on.
  • the time step is to be selected as small as possible depending on the computing time and / or information gain, eg one second.
  • the set of collective maneuvers includes possible collective maneuvers, preferably all possible collective maneuvers, in one time step.
  • a collective maneuver from the set of collective maneuvers can be defined as a sequence of continuous formations from the start formation to a final formation.
  • Each formation of a collective maneuver fulfills a given set of conditions, pairwise, lateral relationships and / or one Passing sequences specify one or more critical areas of the traffic scenario.
  • a critical area may be an area of a road section that may not be traveled during a collective maneuver of a vehicle at any particular time or at all.
  • a collective maneuver from the set of collective maneuvers may include the ego vehicle and one or more other vehicles and / or obstacle objects, and define movement of the ego vehicle relative to at least one other vehicle and / or obstacle object and relative to other vehicles.
  • the set of collective maneuvers may be accomplished by iteratively extending a tree data structure, where each node of the tree data structure corresponds to a formation. Through the iterative extension of the tree data structure possible maneuvers can be searched starting from the current formation. Further, by iteratively extending the tree data structure, a possible maneuver for the ego vehicle can be quickly found, even if the tree data structure has not been fully expanded.
  • the current formation can be used as the root element of the tree data structure.
  • the iterative extension of the tree data structure can be used to determine end formations and intermediate formations between the current formation as the start formation and the end formation (s).
  • An end formation may be an intermediate formation of another path from the current formation to another end formation. This can be used to find maneuvers that are part maneuvers of another maneuver and increase the number of maneuvers in the set of maneuvers from which the ego vehicle can select or determine a maneuver to move.
  • a new formation can be generated as a function of a movement, in particular a relative movement, of one or more vehicles, eg a movement of the ego vehicle and / or a movement of the at least one further vehicle.
  • the predetermined amount of discrete actions may limit a search space in which to search for a possible maneuver.
  • An action on the set of actions describes a change in a relative order of objects, eg, a relative order of vehicles and / or obstacles, of a formation.
  • a vehicle action pair (V, e) e V x A can transform a formation, eg the current formation, into a new formation.
  • the object which is the next object in the traveling direction with respect to the longitudinal order is referred to as the preceding object in the following regardless of the lane.
  • the lateral actions a e ft and a r i 9 ht represent a lane change from the right lane to the left lane or from the left lane to the right lane.
  • the longitudinal action aig ong may represent two different actions: first, passing the preceding object, the passing resulting in a change of position with respect to the longitudinal order of a formation or, second, passing an intersection and placing the vehicle in a lane linked to the intersection.
  • Tree data structure has been fully extended with all possible actions, all possible formations are based on the current formation.
  • clipping conditions may be defined that discard one or more paths of the tree data structure, and / or limit a depth of one or more paths of the tree data structure.
  • the cut-off conditions for example, loops in the generation of new formations can be prevented.
  • Clipping conditions may additionally take into account vehicle dynamics parameters.
  • the following clipping conditions can be defined:
  • Obstacle object on the same track - Inappropriate lane change: Change to the lane of the vehicle ahead when moving to another vehicle; Performing a lane change to a non-existent lane.
  • conditions can be defined that have to fulfill a final formation. For example, it can be determined as a condition for a final formation that all vehicles with different directions of travel must have passed each other, and / or that all vehicles are on a lane and have their correct direction of travel.
  • An end formation may be an intermediate formation for another end formation. This means that extending a path of the
  • Tree data structure can be continued in a final formation until at least one of the cut conditions and / or all final conditions are met.
  • a path from the current formation to a final formation is a maneuver as defined above. Similar
  • Maneuvers i. Maneuvers that have similar paths, but whose result is similar in terms of the respective final formation, can be removed from the set of maneuvers to further reduce the number of maneuvers and thus the complexity of the calculation.
  • FIG. 3 shows an iteratively expanded tree data structure and by the extension of FIG
  • the path from the starting formation F 0 to the final formation FTI specifies a first collective maneuver Mi
  • the path from the starting formation F 0 to the final formation F 2 specifies a second collective maneuver M 2 specifying the path from the starting formation F 0 to the final formation FT3 a third collective maneuver M 3 .
  • the vehicle A follows the vehicle B.
  • the vehicle A overtakes the vehicle B before the vehicle C passes the vehicle B.
  • the vehicle A overtakes the vehicle B after the vehicle C has passed the vehicle B.
  • the method 100 may, after determining 106 the set of collective maneuvers, calculate at least one trajectory per collective maneuver on the set of collective maneuvers 108.
  • Trajectories of a collective maneuver are preferably homotopic, ie one trajectory of the collective maneuver may be in another trajectory of the collective maneuver
  • Maneuvers are continuously transformed in compliance with the given, structural
  • each collective maneuver is calculated from the set of collective maneuvers.
  • the calculated trajectories of a collective maneuver make it possible to compare the collective maneuver with another collective maneuver and with observations of actual trajectories.
  • the calculated trajectories represent a collective maneuver.
  • an optimization problem can be formulated which minimizes a cost function under given conditions.
  • the calculated trajectories are preferably minimum cost trajectories for the collective maneuver.
  • the predetermined conditions may include maneuver-dependent and / or maneuver-independent conditions.
  • Maneuver-independent conditions may include one or more driving dynamics parameters, for example, depending on the or
  • Vehicles can be specified.
  • Maneuver dependent parameters may be spatial parameters, e.g. on which side two vehicles pass the respective other vehicle and / or which safety distance is to be observed with respect to one or more vehicles and / or obstacle objects, and / or time-based parameters, e.g. in which order two or more vehicles pass a critical area.
  • spatial parameters e.g. on which side two vehicles pass the respective other vehicle and / or which safety distance is to be observed with respect to one or more vehicles and / or obstacle objects
  • time-based parameters e.g. in which order two or more vehicles pass a critical area.
  • the conditions are to be determined so as to avoid a collision of two or more vehicles and / or a collision of a vehicle with an obstacle object.
  • the optimization problem can be split into at least two subproblems: an optimization problem for an optimization of a longitudinal control of the ego vehicle and an optimization problem for an optimization of a lateral control of the ego vehicle.
  • the optimization problem for the optimization of the longitudinal control is one
  • the longitudinal optimization is first performed and the results of the longitudinal optimization used as an input for the lateral optimization. This has the advantage that the lateral optimization of the longitudinal positions of the vehicles and / or obstacle objects are known.
  • a quadratic cost function can be chosen.
  • an acceleration, a deviation from a desired speed, and / or a deviation from a lateral position can be included in the calculation of the trajectory (s).
  • Cost function costs of all vehicles and / or obstacles are taken into account.
  • FIG. 4 shows a longitudinal optimization of the traffic scenario from FIG. 2 and the determined collective maneuvers Mi, M 2 , and M 3 from FIG. 3
  • FIG. 5 shows a lateral optimization of the traffic scenario from FIG. 2 and the determined, collective ones Maneuver Mi, M 2 , and M 3 from FIG. 3.
  • FIG. 6 shows the cost-optimal or cost-minimal trajectories for the determined collective maneuvers Mi, M 2 , and M 3 .
  • 402 shows the longitudinal trajectory of the ego vehicle 202, 404 the longitudinal trajectory of the stationary vehicle 204, and 406 the longitudinal trajectory of the further vehicle 206.
  • Regions that are not allowed to cross the longitudinal trajectory 402 of the ego vehicle are indicated 408.
  • Areas that the longitudinal trajectory 406 of the further vehicle 206 is not allowed to cross are marked 410.
  • the regions 408 and / or 410 may be dynamically adjusted between the time steps to accommodate movements of the vehicles, e.g. of the ego vehicle 202 and the further vehicle 206.
  • the lateral trajectory of the ego vehicle 202 with 510, the lateral trajectory of the stationary vehicle 204 with 512, and the lateral trajectory of the further vehicle 206 with 504 marked. Areas that the lateral trajectory of the ego vehicle 202 and the further vehicle 206 may not intersect are 502, areas that are not allowed to cross the lateral trajectory of the ego vehicle are 508, and areas that do not cross the lateral trajectory of the other vehicle may, marked with 506.
  • Trajectories for the ego vehicle 202, the stationary vehicle 204 and the other vehicle 206 for the respective collective maneuvers Mi, M 2 , and M 3 are derived.
  • the trajectories shown in FIG. 6 can be driven directly by the ego vehicle 202.
  • the method 100 may further determine the maneuver of the ego vehicle based on the calculated trajectory (s) for the collective maneuvers from the set of collective maneuvers and a movement of the at least one other vehicle sensed by a sensor of the ego vehicle 1 1 0. Es exactly one collective maneuver is determined from the set of collective maneuvers in one time step. In a subsequent time step, another collective maneuver can be determined. Changing the collective maneuver between two time steps enables the mapping of a behavior of a human driver who continuously assesses a traffic scenario and flexibly changes an intended maneuver, e.g. a change in the traffic scenario occurs or a re-evaluation of the traffic scenario leads to another maneuver.
  • the movement of the ego vehicle can be modeled as a stochastic process comprising all collective maneuvers from the set of collective maneuvers, in a time step only a collective maneuver out of the crowd of collective maneuvers can be active.
  • a maneuver change probability can be defined.
  • the maneuvering likelihood between two time steps for all collective maneuvers from the set of collective maneuvers is 10%.
  • the collective maneuver can be estimated.
  • a Kalman filter e.g. an interactive, multiple model, IMM for short, Kaiman filters can be used to estimate the collective maneuver or states of the collective maneuver.
  • Bayesian Bayesian
  • Fig. 7 shows the calculated probabilities for the collective maneuvers Mi, M 2 , and M 3 .
  • the collective maneuver M 3 has the highest probability. The ego vehicle thus becomes the collective maneuver M 3 in the
  • Obstacle objects are efficiently considered in determining a collective maneuver of vehicles. Finding possible collective maneuvers can be efficiently automated through the use of formations. An ego vehicle can thus more accurately estimate traffic situations if the collective maneuver does not include the ego vehicle. And, the calculated trajectory for the particular collective maneuver can be driven directly by the ego vehicle if the ego vehicle is a vehicle of collective maneuver. The collective maneuver of the ego vehicle can thus be re-determined efficiently with regard to the required computing resources for each time step. In addition, possible, collective maneuvers of the ego vehicle depending on a current
  • the ego vehicle 202 may thus respond efficiently and flexibly to different and rapidly changing traffic scenarios similar to a human driver, in which the ego vehicle 202 makes a new assessment of the traffic scenario at each time step in which the method is performed. In doing so, the ego vehicle can efficiently take into account relative movements between moving vehicles and relative movements to static obstacle objects and quickly determine a collective maneuver to be driven. LIST OF REFERENCE NUMBERS

Abstract

L'invention concerne un procédé (100) pour la détermination d'une manœuvre collective d'au moins deux véhicules, le procédé comprenant : la réception (102) d'un état d'un premier véhicule et d'un état d'au moins un autre véhicule dans le champ d'un ego-véhicule ; la détermination (104) d'une formation actuelle pour le premier véhicule et l'au moins un autre véhicule sur la base de l'état reçu du premier véhicule et de l'état reçu de l'au moins un autre véhicule dans le champ de l'ego-véhicule, la formation actuelle comprenant un ordre relatif du premier véhicule et de l'au moins un autre véhicule par rapport à une voie ; la détermination (106) d'une pluralité de manœuvres collectives basées sur la formation actuelle, une manœuvre collective de la pluralité de manœuvres collectives comprenant une séquence de formations de la formation actuelle jusqu'à une formation finale ; le calcul (108) d'une trajectoire pour le premier véhicule et d'une trajectoire pour l'au moins un autre véhicule pour une manœuvre collective de la pluralité de manœuvres collectives ; et la détermination (110) de la manœuvre collective de la pluralité de manœuvres collectives sur la base des trajectoires calculées de la manœuvre collective de la pluralité de manœuvres collectives et d'un mouvement du premier véhicule et de l'au moins un autre véhicule détecté par un système de capteurs de l'ego-véhicule.
PCT/EP2018/055452 2017-04-26 2018-03-06 Procédé, produit de programme informatique, support lisible par ordinateur, appareil de commande et véhicule comprenant l'appareil de commande pour la détermination d'une manœuvre collective d'au moins deux véhicules WO2018197083A1 (fr)

Priority Applications (3)

Application Number Priority Date Filing Date Title
EP18710403.9A EP3616184A1 (fr) 2017-04-26 2018-03-06 Procédé, produit de programme informatique, support lisible par ordinateur, appareil de commande et véhicule comprenant l'appareil de commande pour la détermination d'une man uvre collective d'au moins deux véhicules
CN201880009506.5A CN110268457B (zh) 2017-04-26 2018-03-06 用于确定至少两个车辆的集体操纵的方法、计算机程序产品、计算机能读取的介质、控制器和包括该控制器的车辆
US16/657,025 US20200050214A1 (en) 2017-04-26 2019-10-18 Method, Computer Program Product, Computer-Readable Medium, Control Unit, and Vehicle Comprising the Control Unit for Determining a Collective Maneuver of at Least Two Vehicles

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
DE102017206987.2 2017-04-26
DE102017206987.2A DE102017206987A1 (de) 2017-04-26 2017-04-26 Verfahren, Computerprogrammprodukt, Computer-lesbares Medium, Steuergerät und Fahrzeug umfassen das Steuergerät zum Bestimmen eines kollektiven Manövers von wenigstens zwei Fahrzeugen

Related Child Applications (1)

Application Number Title Priority Date Filing Date
US16/657,025 Continuation US20200050214A1 (en) 2017-04-26 2019-10-18 Method, Computer Program Product, Computer-Readable Medium, Control Unit, and Vehicle Comprising the Control Unit for Determining a Collective Maneuver of at Least Two Vehicles

Publications (1)

Publication Number Publication Date
WO2018197083A1 true WO2018197083A1 (fr) 2018-11-01

Family

ID=61622557

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/EP2018/055452 WO2018197083A1 (fr) 2017-04-26 2018-03-06 Procédé, produit de programme informatique, support lisible par ordinateur, appareil de commande et véhicule comprenant l'appareil de commande pour la détermination d'une manœuvre collective d'au moins deux véhicules

Country Status (5)

Country Link
US (1) US20200050214A1 (fr)
EP (1) EP3616184A1 (fr)
CN (1) CN110268457B (fr)
DE (1) DE102017206987A1 (fr)
WO (1) WO2018197083A1 (fr)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US11462099B2 (en) 2019-02-08 2022-10-04 Zf Automotive Germany Gmbh Control system and control method for interaction-based long-term determination of trajectories for motor vehicles

Families Citing this family (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US11718300B2 (en) * 2019-02-27 2023-08-08 Zf Automotive Germany Gmbh Method and control unit for a system for controlling a motor vehicle
US20220185290A1 (en) * 2019-04-29 2022-06-16 Volvo Truck Corporation A method for a lane change of a plurality of vehicles
JP7194640B2 (ja) * 2019-05-16 2022-12-22 本田技研工業株式会社 車両制御装置、車両制御方法、およびプログラム
CN112286049A (zh) * 2019-07-27 2021-01-29 华为技术有限公司 运动轨迹预测方法和装置
US11537127B2 (en) * 2019-09-12 2022-12-27 Uatc, Llc Systems and methods for vehicle motion planning based on uncertainty
DE102019129879A1 (de) 2019-11-06 2021-05-06 Zf Friedrichshafen Ag Verfahren sowie Steuergerät zum Steuern eines Kraftfahrzeugs
GB2594079A (en) * 2020-04-16 2021-10-20 Daimler Ag A method for tracking at least one object in the surroundings of an autonomous motor vehicle with a Frenet frame, as well as an assistance system
DE102020112036A1 (de) 2020-05-05 2021-11-11 Valeo Schalter Und Sensoren Gmbh Fahrerassistenz für ein kraftfahrzeug
US11577733B2 (en) 2020-09-30 2023-02-14 GM Global Technology Operations LLC Efficient road coordinates transformations library
DE102020214347A1 (de) 2020-11-16 2022-05-19 Volkswagen Aktiengesellschaft Bahnplanung eines Fahrzeugs
CN114506323B (zh) * 2022-04-20 2022-07-12 北京主线科技有限公司 编队车辆控制方法、装置、设备及介质

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20080065328A1 (en) * 2006-09-08 2008-03-13 Andreas Eidehall Method and system for collision avoidance
DE102012005245A1 (de) * 2012-03-14 2012-09-20 Daimler Ag Unterstützen eines Fahrers eines Kraftfahrzeugs bei einem Überholvorgang
US20150160653A1 (en) * 2013-12-06 2015-06-11 Elwha Llc Systems and methods for modeling driving behavior of vehicles

Family Cites Families (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8352111B2 (en) * 2009-04-06 2013-01-08 GM Global Technology Operations LLC Platoon vehicle management
JP5206752B2 (ja) * 2010-08-30 2013-06-12 株式会社デンソー 走行環境認識装置
DE102011005844A1 (de) * 2011-03-21 2012-09-27 Bayerische Motoren Werke Aktiengesellschaft Automatische Steuerung eines Fahrzeugs
CN102915465B (zh) * 2012-10-24 2015-01-21 河海大学常州校区 一种基于移动生物刺激神经网络的多机器人联合编队方法
DE102013009252A1 (de) * 2013-06-03 2014-12-04 Trw Automotive Gmbh Steuergerät und Verfahren für eine Notfall-Lenkunterstützungsfunktion
US9165477B2 (en) * 2013-12-06 2015-10-20 Vehicle Data Science Corporation Systems and methods for building road models, driver models, and vehicle models and making predictions therefrom
DE102014211507A1 (de) * 2014-06-16 2015-12-17 Volkswagen Aktiengesellschaft Verfahren für ein Fahrerassistenzsystem eines Fahrzeugs
CN104567891A (zh) * 2015-01-22 2015-04-29 北京微车一族信息技术发展有限公司 一种车队行车的地图导航方法
US9821801B2 (en) * 2015-06-29 2017-11-21 Mitsubishi Electric Research Laboratories, Inc. System and method for controlling semi-autonomous vehicles
DE102015217891A1 (de) * 2015-09-17 2017-03-23 Volkswagen Aktiengesellschaft Bestimmen einer Soll-Trajektorie für ein Fahrzeug

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20080065328A1 (en) * 2006-09-08 2008-03-13 Andreas Eidehall Method and system for collision avoidance
DE102012005245A1 (de) * 2012-03-14 2012-09-20 Daimler Ag Unterstützen eines Fahrers eines Kraftfahrzeugs bei einem Überholvorgang
US20150160653A1 (en) * 2013-12-06 2015-06-11 Elwha Llc Systems and methods for modeling driving behavior of vehicles

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US11462099B2 (en) 2019-02-08 2022-10-04 Zf Automotive Germany Gmbh Control system and control method for interaction-based long-term determination of trajectories for motor vehicles

Also Published As

Publication number Publication date
EP3616184A1 (fr) 2020-03-04
CN110268457B (zh) 2022-08-23
DE102017206987A1 (de) 2018-10-31
CN110268457A (zh) 2019-09-20
US20200050214A1 (en) 2020-02-13

Similar Documents

Publication Publication Date Title
WO2018197083A1 (fr) Procédé, produit de programme informatique, support lisible par ordinateur, appareil de commande et véhicule comprenant l'appareil de commande pour la détermination d'une manœuvre collective d'au moins deux véhicules
DE102016107705B4 (de) Verfahren zur adaptiven erzeugung eines geplanten pfads für ein manöver eines autonomen fahrens
DE112017003517T5 (de) Steuergerät und Verfahren zur Steuerung eines Fahrzeugs und nichtflüchtiger computerlesbarer Speicher
EP2981951B1 (fr) Parcours automatique d'un trajet
DE102019108477A1 (de) Automatische navigation unter verwendung von deep reinforcement learning
DE102019201124A1 (de) Ein System für ein Fahrzeug
DE112017000787T5 (de) Verfahren zum Steuern der Bewegung eines Fahrzeugs und Fahrzeugsteuersystem
DE102017119952A1 (de) Autonomes fahrzeug mit kursprädiktion
EP3144920A1 (fr) Détermination d'une trajectoire théorique d'un véhicule
DE102016212700A1 (de) Verfahren und System zur Steuerung eines Fahrzeugs
DE102015208790A1 (de) Bestimmen einer Trajektorie für ein Fahrzeug
DE102019118366A1 (de) Verfahren sowie Steuergerät für ein System zum Steuern eines Kraftfahrzeugs
DE102020211017B3 (de) Zuordnung von Ampeln zu zugehörigen Fahrspuren
DE102016007899B4 (de) Verfahren zum Betreiben einer Einrichtung zur Verkehrssituationsanalyse, Kraftfahrzeug und Datenverarbeitungseinrichtung
DE102018217845A1 (de) Einrichtung und Verfahren zum Regeln eines Prozesses
DE102018213971A1 (de) Verfahren und Vorrichtung zur Auswahl eines Fahrmanövers
WO2020177958A1 (fr) Procédé et unité de commande pour reconnaître un véhicule changeant de file
EP3898370A1 (fr) Procédé et système de commande d'un véhicule automobile
DE102018204101A1 (de) Verfahren und Vorrichtung zur Planung einer Trajektorie für die Längs- und/oder Querführung eines Fahrzeugs
DE102020127855A1 (de) Sicherheitssystem, automatisiertes fahrsystem und verfahren dafür
DE102021105159A1 (de) Erfassung von fahrzeugen und reaktion darauf
DE102019133970A1 (de) Steuern eines Ego-Fahrzeugs in einer Umgebung einer Vielzahl von Durchfahrten
DE102017200580A1 (de) Verfahren zur Optimierung einer Manöverplanung für autonom fahrende Fahrzeuge
DE102020117767A1 (de) Trajektorienplanung für ein autonomes Fahrmanöver eines Ego-Fahrzeugs
DE102019122250A1 (de) Verfahren sowie Steuergerät für ein System zum Steuern eines Kraftfahrzeugs

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 18710403

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

ENP Entry into the national phase

Ref document number: 2018710403

Country of ref document: EP

Effective date: 20191126